DevOps SRE
Post #7303 · source on Telegram
AWS Fired 40% of DevOps Team Then Let AI Take Their Jobs Headline
Description
A screenshot of a Medium article headline reading 'AWS Just Fired 40% of Its DevOps Team -- Then Let AI Take Their Jobs!' with the subheading 'Leaked internal tools show how Amazon's cloud is now self-healing, self-scaling, and self-negotiating -- no humans required.' The article is by Mohab AbdelKarim, marked as a 3 min read from 2 days ago. The '2 days ago' timestamp is highlighted with a red box, possibly to emphasize the recency or dubious nature of the clickbait claim. This represents the growing anxiety about AI replacing DevOps/SRE roles
Use J and K for navigation
Comments
20Comment deleted
Self-healing, self-scaling, self-negotiating -- so basically AWS built an infrastructure that can do everything except explain its own pricing model
The new AWS AI is great at self-healing and self-scaling, but it just submitted a pull request to replace the entire billing system with a single, infinitely recursive Lambda function. Turns out 'self-negotiating' also applies to its own budget
Sure, the cloud is now self-healing and self-negotiating - until 2 a.m. when the AI decides the cheapest way to resolve an incident is to page the one remaining human… who wrote the on-call bot in the first place
The real joke is that the AI will spend 40% of its compute cycles trying to figure out why the previous DevOps team had 17 different ways to deploy the same microservice, each with its own undocumented quirks that somehow kept production running
Finally, AWS achieved the ultimate DevOps dream: zero human intervention. Turns out the 'you build it, you run it' philosophy works even better when 'you' is an AI that never sleeps, never complains about on-call, and definitely won't ask for equity refreshers
Call me when their self‑negotiating cloud can win an Enterprise Discount Agreement; until then it’s just an expensive cron job with a press release
Peak SRE: Automating toil until the infra fires you for being the new single point of failure
A disaster waiting to happen. Comment deleted
Sourceless article on Medium Comment deleted
Article seems to be here: https://blog.stackademic.com/aws-just-fired-40-of-its-devops-team-then-let-ai-take-their-jobs-d9db9d298bfa The layoffs were in July: https://www.reuters.com/business/retail-consumer/amazons-aws-cloud-computing-unit-cuts-least-hundreds-jobs-sources-say-2025-07-17/ Comment deleted
But, various comments here and there suggesting the ai opentofu stuff is not real? Comment deleted
https://archive.is/b6aUD#selection-641.1-643.17 Paywall bypass link, seems entirely shite tbqh. Doubt any of that exists Comment deleted
Here's an older article, from after the actual layoffs, citing the same bullshit: https://aws.plainenglish.io/aws-laid-off-40-of-its-devops-staff-what-theyre-using-instead-will-shock-you-544ebb38a63d Paywall bypass: https://archive.is/bfst8 Also seems likely to be rubbish Comment deleted
Even if it was to replace them with other humans would be a disaster Comment deleted
Manager: Fix the disaster. AI: Done. Manager: Looks like it’s not fixed. AI: You’re right. Sorry - my bad. Now it’s fixed. Manager: Still not fixed. AI: You’re right. Sorry - my bad. Now it’s definitely fixed. Manager: F*ck Comment deleted
The manager just will add another AI to fix it faster Comment deleted
The issue isn’t speed - it's that the AI hallucinates and lies that it fixed the problem Comment deleted
But did you provide it with tools and flow to verify issue being actually fixed? 🤔 Comment deleted
Yes. I explicitly ask it to verify. Some times it prefer to straight up lie. Claude Code opus 4.1 Comment deleted
But who would tell the manager? Comment deleted